基于马尔科夫决策过程理论的无线网络流媒体传输研究

发布时间：2018-09-12 06:20

【摘要】：当今社会人们对视频传输的要求越来越高,随着技术的发展,基于HTTP(Hypertext Transfer Protocol)的动态自适应流媒体技术越来越受到人们的关注。基于HTTP的动态自适应视频流传输(DASH:Dynamic Adaptive Streaming based on HTTP)的技术原理是把相同的视频编码为多份不同质量的视频文件,然后把这些文件分为很多的视频分段存储在服务器端,在流媒体传输过程中,客户端可以根据当前带宽变化来选择适应此网络带宽的质量分段进行传输。但是,DASH标准并没有规定客户端选择视频文件分段时所应采取的策略,这使得客户端可以决定采取什么策略来进行高效的传输,特别是针对时变性很强的无线信道而言。客户端如何根据时变的无线网络带宽选择合适的视频质量分段成为了当今流媒体传输技术的一个研究热点。本文在前人工作的基础上,进一步研究了基于马尔科夫决策过程理论的流媒体传输系统模型,基于模型求解得到最优传输策略。重点讨论和分析了在自适应传输过程中,分段质量和播放时限是如何影响最佳策略的,并分析了在马尔科夫决策过程理论模型中,下载时间离散化对系统模型准确度的影响。本文研究的主要内容为:1.参考国内外已有研究工作,对马尔科夫决策过程理论(MDP)进行学习与总结,进一步研究了流媒体传输中的马尔科夫决策过程系统建模。2.基于该模型,求解得到无线网络流媒体传输过程中的最优传输策略,即根据当前系统状态选择最优的视频分段质量,以期达到最好的用户观看体验。3.基于最优策略,设计仿真程序进行仿真,验证了最优策略的有效性,并通过仿真分析了系统模型在计算中存在的误差因素,即系统状态离散化带来的误差。
[Abstract]:With the development of technology, the dynamic adaptive streaming media technology based on HTTP (Hypertext Transfer Protocol) has attracted more and more attention. Dynamic Adaptive Video Stream Transmission based on HTTP (the principle of DASH:Dynamic Adaptive Streaming based on HTTP) is to encode the same video into multiple video files of different quality, and then divide these files into many video segments and store them on the server side. In the process of streaming media transmission, the client can select the quality segment which adapts to the network bandwidth according to the current bandwidth change. However, the DASH standard does not specify the strategy to be adopted when the client selects video file segmentation, which enables the client to decide what strategy to adopt for efficient transmission, especially for the time-varying wireless channel. How to select the appropriate video quality segment according to the time-varying wireless network bandwidth has become a hot topic in streaming media transmission technology. In this paper, based on the previous work, the model of streaming media transmission system based on Markov decision process theory is further studied, and the optimal transmission strategy is obtained based on the solution of the model. This paper mainly discusses and analyzes how segment quality and playback time affect the optimal strategy in adaptive transmission, and analyzes the influence of download time discretization on the accuracy of system model in Markov decision process theory model. The main contents of this paper are: 1: 1. Referring to the existing research work at home and abroad, the Markov decision process theory (MDP) is studied and summarized, and the Markov decision process modeling in streaming media transmission is further studied. Based on this model, the optimal transmission strategy in the process of streaming media transmission in wireless network is solved, that is, the optimal video segment quality is selected according to the current system state, in order to achieve the best user viewing experience. 3. Based on the optimal strategy, a simulation program is designed to verify the effectiveness of the optimal strategy, and the error factors in the calculation of the system model are analyzed by simulation, that is, the error caused by the discretization of the system state.
【学位授予单位】：西华师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.8

【参考文献】